QCM-2013-2-0007.Performance of Time Dependent Density Functional Theory in the strong field photoionisation of noble gas atoms


Status: submitted project

The main idea behind this project is to test the performance of Time Dependent Density Functional Theory (TDDFT) in the strong high intensity regime for two noble gas atoms, Neon and Argon (we still have not tested Xenon because we are still testing Argon), with respect to the lowest order non vanishing perturbation theory (LOPT) results that have been obtained previously. In order to do this, I have used their same parameters and I have performed the calculations with the Octopus code. In particular I have compared the number of ejected electrons obtained from both methods. In order to simulate our systems I have placed them inside a box with a certain dimension limiting our systems in space. A limited box in space leads to reflections of the outgoing electrons bouncing back to the system which is unphysical. Therefore a relatively big box is necessary to converge our calculations. For Neon we have seen that our TDDFT results are good but this hasn't been the case for Argon, which is why we want to extend our application period. For Argon apart from the tests that I have performed mentioned below in the Research Project Description area, we have performed other tests to understand why this was the case and that is why all the assigned Marenostrum hours for this period have already been used. Note that as explained below, for each new pseudopotential, absorbing boundary method tested ... we have to test two pulse lengths per atom, 7 intensities per functional and 8 different functionals... which leads to many calculations.

From our calculations, we have reached the following conclusions:

Argon is stabler than Neon in time with respect to the ejection of electrons.

I have computed the cross sections and compared them to the experimental ones to show that for both Ne and Ar we need a box of 30 atomic units to avoid unphysical reflections. For both cases the theoretical cross section underestimates the experimental one. Therefore we cannot understand why we get better results for Neon than for Argon from the cross sections.

The results hardly change if I use different absorbing methods but they do improve if we use the 105eV
laser pulse and they particularly improve if we use a harder pseudopotential. This might be why we
obtain better results for Neon, as for Neon we use a 93eV laser and it is more strongly bound because of its size, so that the number of escaped electrons is smaller than for Argon. Therefore for Argon we
either need a harder two inner core and eight semi inner core pseudopotential or a softer two inner
core only pseudopotential which is tougher computationally because more memory is required.

As the pseudopotential becomes harder, the ground state eigenvalues tend towards the all-electron eigenvalue limit. We are testing the transferability of the pseudopotential, because the pseudopotentials have been generated for the neutral configuration.

If the electrons do not interact, the results get worse, so that the electron interaction effects are not negligible.

To summarise, we want to test now the effect of the two new different pseudopotentials: a harder two inner core and eight semi inner core pseudopotential AND a softer two inner
core only pseudopotential. We want to test them using a box size of 30 atomic units (because we avoid box states by doing so) and these require a very small spacing to achieve convergence.
For each pseudopotential, note that we want two pulse lengths per atom, 7 intensities per functional and 8 different functionals. I would also like to install a new octopus version which outputs the ionisation channels. This would be useful to compare how the LOPT ones compare to our TDDFT ones. This option can be added when I submit my calculations, but due to this new extra feature which has to be outputted, the calculations will also take longer.


Team Leader: Angel Rubio Secades
Alison Crawford Uranga
Esa Räsänen
Umberto De Giovannini
Micael J. T. Oliveira
Peter Lambropoulous
Stefan Kurth

Related Research Areas